互联网 qkzz.net
全刊杂志网:首页 > 女性 > 文章正文
刊社推荐

一种基于灰色关联分析和超效率DEA的MCDM模型李 凌 刘建永 付成群 伍中军


摘 要:针对多属性决策问题(MCDM)权重难以客观确定的缺陷,提出一种基于灰色关联分析和超效率数据包络分析(DEA)的混合算法研究MCDM。对MCDM进行建模,利用灰色关联分析计算各属性的点关联度,为了得到各属性的相对最优灰关联度并将其充分排序,利用超效率DEA改进灰色关联分析模型。该混合算法能够更加客观地确定权重向量,并且突破权重和为1的限制,使得均一的权重更加灵活;同时,可以得到各属性的相对最优灰色关联度,超效率DEA能够增强模型对灰关联度的分辨能力。通过北京市商品房空置影响因素的实例分析验证了该混合算法的有效性。
  关键词:灰色关联分析; 超效率数据包络分析; 权重; 多属性决策; 空置商品房
  中图分类号:TP39
  文献标志码:A
  
  文章编号:1001-3695(2010)02-0526-03
  doi:10.3969/j.issn.1001-3695.2010.02.034
  
  MCDM model based on gray correlation analysis andsuper-efficiency DEA
  
  LI Ling1, LIU Jian-yong1, FU Cheng-qun1, WU Zhong-jun1,2
  
  (1. Engineering Institute of Engineering Corps, PLA University of Science & Technology, Nanjing 210007, China; 2. Command & Engineering Institute of Chemical Defense, Beijing 102205, China)
  
  Abstract:In view of the defect that the weights of multiple criteria decision making (MCDM) are hard to determine objectively, this paper proposed a hybrid model combined gray correlation analysis with super-efficiency DEA (data envelopment analysis). On the basis of gray correlation analysis, it calulated the point correlation degree of each criterion. In order to obtain relative optimal gray correlation degree of each criterion and rank them completely, suggested the super-efficiency DEA method. The hybrid model could ascertain the weight vector more objectively. It broke through the constraint that the sum of each weight vector was equal to 1, thus the homogeneous weight could be more flexible. The improved model could lead to relative optimal gray correlation degree of each criterion. Super-efficiency DEA could enhance the discernment ability as well. It stu-died the influence factors of vacant commercial housing in Beijing city.The results show that the correlation order of influence factors. The numerical example illustrates the availability and efficiency of the hybrid model. ......
很抱歉,暂无全文,若需要阅读全文或喜欢本刊物请联系《计算机应用研究》杂志社购买。
欢迎作者提供全文,请点击编辑
分享:
 

了解更多资讯,请关注“木兰百花园”
分享:
 
精彩图文


关键字
支持中国杂志产业发展,请购买、订阅纸质杂志,欢迎杂志社提供过刊、样刊及电子版。
关于我们 | 网站声明 | 刊社管理 | 网站地图 | 联系方式 | 中图分类法 | RSS 2.0订阅 | IP查询
全刊杂志赏析网 2017